High Quality Reconstruction Of Cone-Beam And Fan-Beam Computed Tomography

X-ray computed tomography (CT) has experienced tremendous growth in recently 30 years and has achieved substantial enhancement both in efficiency and precise. Since the introduction of helical and multi-slice CT, many new clinical applications have been developed. CT is again becoming one of the most exciting modalities in the medical imaging field.In 1980, a novel 3-D reconstruction method was proposed, which plays a critically important role in the history of CT imaging although that was an approximate method. Because FDK is satisfied with small cone angular high quality imaging, it is the main stream algorithm for commercial applications. But with the development of multi-row detector and the improvement of imaging speed and precise in modern clinical application, fast and precise 3-D cone beam CT reconstruction method study presents the highlight project for many research departments.Modern clinical techniques require as less as possible radiation towards normal tissues surrounding the diseased region, which would reduce unnecessary harm to the healthy organs. At the same time, we expect to get better reconstructed images with low dose radiation. Therefore, one side we can use a special region of interest, such as only diseased region imaging, which required to accommodate incomplete projection data with serious truncation. Other side we can conduct to understand the relationship between the noise present in the image and the outcome of diagnosis, which supervise the novel low dose CT reconstruction.More considerably, high-density objects, such as metallic implants, surgical clips and dental fillings, often cause severe artifacts on computed tomography scans, and many render images non-diagnostic. This problem is due to the fact that satisfactory images cannot be calculated from projections if data are missing or distorted. So, metal artifacts reduction is also the highlight for the CT application research, which is helpful for the radiotherapy process.The main contributions in this PhD dissertation are as following,We present an alternative approach to reconstructing exactly an image from helical cone-beam projections, which is the combination of Hilbert filter and Ramp filter. Based on the Katsevich reconstructed algorithm framework, the proposed algorithm takes the different advantages of the FDK-type algorithms and Katsevich algorithm, which completely avoids the direct derivatives with respect to the coordinates on the detector plane. This alternative expression has a significant practical implication, thus leading to the images with quality improvement and reduced artefacts. At the same time, although the filtering process of the proposed algorithm is composed of Hilbert and Ramp filters, in contrast the FDK-type reconstructed algorithm, however, the new still performs a 1D shift-invariant filtering of the modified data on the detector plane and the redundancy weight is applied after filtering, allowing a more efficient numerical implementation. The complexity of the new algorithm compared with Katsevich algorithm does not extra increase.We present a new FBP image reconstruction algorithm based on the Katsevich's original algorithm paradigm. This proposed algorithm successfully avoids the direct derivatives with respect to rotation angle, which achieves good image quality improvement and fewer artifacts. The new algorithm still performs a 1D shift-invariant filtering of the modified data on the detector plane and the redundancy weight is applied after filtering, allowing a more efficient numerical implementation. Results in these studies confirm the observation that the proposed algorithm can improve the image resolution over Katsevich's original algorithm with noiseless and noise projection data.An improved super-short-scan reconstruction algorithm was proposed for fan-beam computed tomography based on Pi-line. The new algorithm was also based on a FBP algorithm, which could achieve exact region of interest (ROI) recon- struction if and only if all lines passing through the region of interest (ROI) intersect the source trajectory. The new reconstruction formula successfully avoided the direct derivatives with respect to projection data and was expressed as the combination of Hilbert filter and Ramp filter instead of only Hilbert filter, which increased the numerical stability and achieved good image quality improvement in the real discrete data reconstruction.In order to obtain the high quality reconstruction images for low-dose CT, we proposed a new generalized Gibbs prior based low-dose CT reconstruction method. First, CT projection data was modeled as a statistics process. Then based on Bayesian maximum a posteriori estimation method, we designed a novel Gibbs prior named as generalized Gibbs prior, which exploits nonlocal information of the data to suppress noise. Last, we used the filtered back-projection method to finish the final CT reconstruction. The reason for the name of generalized Gibbs prior is that it has been shown to suppress noise effectively while capturing sharp edges without oscillations through the selection of larger neighborhood and adaptively weight form with global information of an image.We develop a corrective method in which the distorted segments in sinogram were identified and interpolated using non distorted neighbor projections, to reduce distorted tomography metal artifacts caused by high-density objects. First, the Anisotropic Gaussian (ASG) Pre-filter reduces the noise content and smoothes streak artifacts in CT image. Next, the filtered image is segmented into several regions by mutual information maximized segmentation (MMS). Then the artifacts class is converted to the CT number with the surrounding material, called "artifact-tissue" class, and after that an "artifact-tissue sinogram" was produced using forward projection method. A final image is reconstructed by the filtered back-projection from appropriately combination of original sinogram and artifacts-tissue sinogram.